{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:RPSMOH7JMBOJ2TKNDT7EBT4QZD","short_pith_number":"pith:RPSMOH7J","schema_version":"1.0","canonical_sha256":"8be4c71fe9605c9d4d4d1cfe40cf90c8f4bf70754070bf13bc66f479b3afa686","source":{"kind":"arxiv","id":"2606.06828","version":1},"attestation_state":"computed","paper":{"title":"AdaGRPO: A Capability-Aware Adaptive Enhancement for Flow-based GRPO","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Dahua Lin, Jiaqi Wang, Jiazi Bu, Pengyang Ling, Tianyi Wei, Tong Wu, Xiaohang Zhan, Xingang Pan, Yibin Wang, Yuhang Zang, Yujie Zhou","submitted_at":"2026-06-05T02:07:08Z","abstract_excerpt":"Group Relative Policy Optimization (GRPO) has demonstrated remarkable success in aligning text-to-image (T2I) flow models with human preferences. However, we have identified that the learning loop of current flow-based GRPO is fundamentally decoupled from the learner's current capability, suffering from critical blind spots at both prompt selection and advantage estimation: (i) Existing methods sample prompts randomly, overlooking the substantial impact of data selection on reinforcement learning (RL) efficacy--a factor proven crucial in GRPO for large language models; (ii) They evaluate sampl"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2606.06828","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-06-05T02:07:08Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"6c715791a4f479a1faf094fdafd25e80229b87446f397e67f8ef76e851817683","abstract_canon_sha256":"32fe19b0e05ec39e37035e5536dd904f2909b1a6be321dffb04f650bea2aa7e9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-08T01:04:30.489764Z","signature_b64":"RZ7fFQE4IGDey/i3wg2bhjpohGOE6fMONRf0C4cL9BJBKWOnA7EgtZ8ebNV2aDNNhn2Adx7xkH0FU7cTiY9CAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8be4c71fe9605c9d4d4d1cfe40cf90c8f4bf70754070bf13bc66f479b3afa686","last_reissued_at":"2026-06-08T01:04:30.488796Z","signature_status":"signed_v1","first_computed_at":"2026-06-08T01:04:30.488796Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"AdaGRPO: A Capability-Aware Adaptive Enhancement for Flow-based GRPO","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Dahua Lin, Jiaqi Wang, Jiazi Bu, Pengyang Ling, Tianyi Wei, Tong Wu, Xiaohang Zhan, Xingang Pan, Yibin Wang, Yuhang Zang, Yujie Zhou","submitted_at":"2026-06-05T02:07:08Z","abstract_excerpt":"Group Relative Policy Optimization (GRPO) has demonstrated remarkable success in aligning text-to-image (T2I) flow models with human preferences. However, we have identified that the learning loop of current flow-based GRPO is fundamentally decoupled from the learner's current capability, suffering from critical blind spots at both prompt selection and advantage estimation: (i) Existing methods sample prompts randomly, overlooking the substantial impact of data selection on reinforcement learning (RL) efficacy--a factor proven crucial in GRPO for large language models; (ii) They evaluate sampl"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.06828","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2606.06828/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2606.06828","created_at":"2026-06-08T01:04:30.488954+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.06828v1","created_at":"2026-06-08T01:04:30.488954+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.06828","created_at":"2026-06-08T01:04:30.488954+00:00"},{"alias_kind":"pith_short_12","alias_value":"RPSMOH7JMBOJ","created_at":"2026-06-08T01:04:30.488954+00:00"},{"alias_kind":"pith_short_16","alias_value":"RPSMOH7JMBOJ2TKN","created_at":"2026-06-08T01:04:30.488954+00:00"},{"alias_kind":"pith_short_8","alias_value":"RPSMOH7J","created_at":"2026-06-08T01:04:30.488954+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RPSMOH7JMBOJ2TKNDT7EBT4QZD","json":"https://pith.science/pith/RPSMOH7JMBOJ2TKNDT7EBT4QZD.json","graph_json":"https://pith.science/api/pith-number/RPSMOH7JMBOJ2TKNDT7EBT4QZD/graph.json","events_json":"https://pith.science/api/pith-number/RPSMOH7JMBOJ2TKNDT7EBT4QZD/events.json","paper":"https://pith.science/paper/RPSMOH7J"},"agent_actions":{"view_html":"https://pith.science/pith/RPSMOH7JMBOJ2TKNDT7EBT4QZD","download_json":"https://pith.science/pith/RPSMOH7JMBOJ2TKNDT7EBT4QZD.json","view_paper":"https://pith.science/paper/RPSMOH7J","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.06828&json=true","fetch_graph":"https://pith.science/api/pith-number/RPSMOH7JMBOJ2TKNDT7EBT4QZD/graph.json","fetch_events":"https://pith.science/api/pith-number/RPSMOH7JMBOJ2TKNDT7EBT4QZD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RPSMOH7JMBOJ2TKNDT7EBT4QZD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RPSMOH7JMBOJ2TKNDT7EBT4QZD/action/storage_attestation","attest_author":"https://pith.science/pith/RPSMOH7JMBOJ2TKNDT7EBT4QZD/action/author_attestation","sign_citation":"https://pith.science/pith/RPSMOH7JMBOJ2TKNDT7EBT4QZD/action/citation_signature","submit_replication":"https://pith.science/pith/RPSMOH7JMBOJ2TKNDT7EBT4QZD/action/replication_record"}},"created_at":"2026-06-08T01:04:30.488954+00:00","updated_at":"2026-06-08T01:04:30.488954+00:00"}